Hortonworks, which sells a distribution of big data framework Apache Hadoop, has released a beta version of its Hortonworks Data Platform (HDP) for Windows Server. The company, which was co-founded by Yahoo, hopes to release a finished version within the next few months. Hadoop is a Java framework for scalable, distributed systems, now developed under the auspices of the Apache Foundation, which uses the MapReduce approach popularised by Google. Until now, Hortonworks and other Hadoop distributors have generally restricted themselves to offering products for Linux.
The product has been developed in partnership with Microsoft, with both companies reported to have worked on the project over the last 18 months; the partnership was first announced in October 2012. At the time, it was stated that Microsoft customers would be offered a Windows Server and Windows Azure-compatible version of Hadoop via HDInsight Server for Windows and the Windows Azure HDInsight service.
In Hortonworks’ view, a Windows version of its platform is necessitated by a large proportion of companies which deploy Windows Server for their applications. The company thus sees a potential for “massive expansion”. The new distribution will not, however, support mixed deployment environments in which both Windows and Linux nodes are installed – users will have to choose either one or the other.
The Windows version should eventually include the same components as the Linux version, but will initially be based on HDP 1.1, rather than the current version 1.2. It will support both Windows Server 2008 and Windows Server 2012. Like its Linux counterpart, the Windows version will be open source. Because Hortonworks Data Platform for Windows and the HDInsight service on Windows Azure are based on the same code base, Microsoft customers should be able to port their Hadoop applications between local and cloud environments without too much difficulty.
Microsoft is not alone in choosing to support a Hadoop distribution rather than launch its own independent product. Oracle has also lined up behind big data framework Hadoop by partnering with distributor Cloudera. SAP has formed partnerships with several different Hadoop distributors.